2.3 Specific heuristics

This list was written by A.H. Schoenfeld:

  • •
    ​

    draw a diagram (if at all possible)

  • •
    ​

    examine special cases

    • –
      ​

      choose special values to exemplify the problem and get a "feel" for it

    • –
      ​

      examine limiting cases to explore the range of possibilities

    • –
      ​

      set any integer parameters equal to 1,2,3,…1,2,3,... in sequences and look for an inductive pattern.

  • •
    ​

    try to simplify the problem

    • –
      ​

      exploiting symmetry

    • –
      ​

      without loss of generality

  • •
    ​

    consider essentially equivalent problems

  • •
    ​

    replace the conditions with equivalent ones

  • •
    ​

    recombine the elements of the problem in different ways

  • •
    ​

    introduce auxiliary elements (e.g. substitutions)

  • •
    ​

    reformulate the problem

  • •
    ​

    change of perspective or notation

  • •
    ​

    considering argument by contradiction or contrapositive

  • •
    ​

    assuming you have a solution, and determining its properties (i.e. what would a valid answer look like)

  • •
    ​

    consider slightly modified problems

    • –
      ​

      choose sub-goals (obtain partial fulfilment of the conditions)

    • –
      ​

      relax a condition and then try to re-impose it

    • –
      ​

      decompose the domain of the problem and work on it case by case

  • •
    ​

    consider broadly modified problems

    • –
      ​

      construct an analogous problem with fewer variables

    • –
      ​

      hold all but one variable fixed to determine that variable’s impact

    • –
      ​

      try to exploit any related problems which have similar

      • *
        ​

        form

      • *
        ​

        "givens"

      • *
        ​

        conclusions

  • •
    ​

    verifying solutions

    • –
      ​

      Does it pass these simple tests?

      • *
        ​

        Does it use all the pertinent data?

      • *
        ​

        Does it conform to reasonable estimates or predictions?

      • *
        ​

        Does it withstand tests of symmetry, dimension analysis, or scaling?

    • –
      ​

      Does it pass these general tests?

      • *
        ​

        Can it be obtained differently?

      • *
        ​

        Can it be substantiated by special cases?

      • *
        ​

        Can it be reduced to known results?

      • *
        ​

        Can it be used to generate something you know?